The basic helix-loop-helix transcription factor Mist1 functions as a transcriptional repressor of myoD.

Mist1 also inhibits MyoD from inducing muscle fiber formation in transfectedC3H10T1/2 fibroblasts and represses MyoD from activating gene transcription inmyogenic cells.. In contrast, at

EMBO J 1998 March 2; 17(5): 1412–1422.

The basic helix-loop-helix transcription factor Mist1 functions

as a transcriptional repressor of myoD.

Claudie Lemercier1,2, Robert Q To1,3, Rosa A Carrasco, and Stephen F Konieczny*

1 Department of Biological Sciences, Purdue University, West Lafayette, IN 47907-1392, USA

2 Present address: Laboratoire de Greffe de Moelle, UMR 5540, Universite Bordeaux 2, 146 rue Leo Saignat, 33076 Bordeaux Cedex, France

3 Present address: The Veterans Affairs Medical Center, 4150 Clement Street, San Francisco, CA 94121, USA

* Corresponding author: Stephen F Konieczny, sfk@bilbo.bio.purdue.edu

Received 23 October 1997; Revised 19 December 1997; Accepted 19 December 1997

Abstract

A good model system to examine aspects of positive and negative transcriptionalregulation is the muscle-specific regulatory factor, MyoD, which is a basic helix-loop-helix(bHLH) transcription factor Although MyoD has the ability to induce skeletal muscleterminal differentiation in a variety of non-muscle cell types, MyoD activity itself is highlyregulated through protein-protein interactions involving several different co-factors Here

we describe the characterization of a novel bHLH protein, Mist1, and how it influencesMyoD function We show that Mist1 accumulates in myogenic stem cells (myoblasts) andthen decreases as myoblasts differentiate into myotubes Mist1 functions as a negativeregulator of MyoD activity, preventing muscle differentiation and the concomitantexpression of muscle-specific genes Mist1-induced inhibition occurs through acombination of mechanisms, including the formation of inactive MyoD-Mist1 heterodimersand occupancy of specific E-box target sites by Mist1 homodimers Mist1 lacks a classictranscription activation domain and instead possesses an N-terminal repressor regioncapable of inhibiting heterologous activators Thus, Mist1 may represent a new class ofrepressor molecules that play a role in controlling the transcriptional activity of MyoD,ensuring that expanding myoblast populations remain undifferentiated during earlyembryonic muscle formation

Introduction

During embryonic development, specific biochemical and morphological changes occur inresponse to extracellular signals that target individual cells within a particular organ ortissue In many cases, the responses represent major changes in gene expressionpatterns To ensure that gene activity is modulated correctly under these conditions, thetranscription machinery must be regulated precisely by a number of positive- andnegative-acting transcription factors Among the different transcription factor familiesthat have been identified, the basic helix–loop–helix (bHLH) family has been shown to be

a key regulator of many different developmental pathways In the case of skeletal muscle,four bHLH muscle regulatory factors (MRFs) (known as MyoD, myogenin, MRF4 and Myf-5)are involved in the differentiation and maintenance of the skeletal muscle phenotype(reviewed in Buckingham, 1994; Olson and Klein, 1994; Ludolph and Konieczny, 1995).Forced expression of any one MRF in non-muscle cells results in the conversion of the

cells to a myogenic stem cell population that can be induced to express genes involved interminal differentiation The MRFs, like most bHLH proteins, form homodimers, butpreferentially heterodimerize with the widely expressed E2A gene products, E12 and E47

(Blackwell and Weintraub, 1990; Lassar et al., 1991), or with the related bHLH protein HEB (Zhang et al., 1991; Hu et al., 1992) It is this high affinity, stable heterodimer

complex that binds to a conserved DNA motif (E-box; -CANNTG-) which is present in thepromoter or enhancer regions of many developmentally regulated genes, including genesinvolved in myogenic differentiation events

The MRFs are classified as myogenic 'activator' proteins because each contains at leastone activation domain that is able to induce gene transcription However, not all bHLHproteins exhibit activator properties Many function instead as negative regulators of

gene transcription For example, Id (Benezra et al., 1990), which possesses a HLH motif

but lacks a basic domain, is capable of forming heterodimers with bHLH proteins and

thereby preventing DNA binding Id inhibits MyoD activity in vivo by forming either

transcriptionally inactive complexes of MyoD–Id or by forming heterodimers with proteins and effectively blocking the formation of active MyoD–E-protein complexes

E-(Benezra et al., 1990; Jen et al., 1992) A second class of negative regulators are the HES

proteins, which are bHLH factors that exhibit a repressor activity that manifests itself

through binding to an E-box-related (N-box) DNA target sequence (Sasai et al., 1992).

Similarly, the mouse bHLH protein Twist has been shown to alter the activity of the MyoDprotein family by blocking DNA binding, by titrating E-proteins and by inhibiting trans-

activation of the MyoD co-factor, MEF2 (Spicer et al., 1996; Hamamori et al., 1997) Thus,

both bHLH and HLH proteins can modulate effectively the transcriptional activity of theMyoD family and regulate directly how MyoD controls myogenic events duringdevelopment

The above studies have revealed the importance of bHLH proteins in regulating both genetranscription and cell fate determination They have also demonstrated that active andnegative strategies are involved in controlling the establishment and maintenance of aparticular phenotype In order to understand better how bHLH proteins interact andregulate each other's activities, we recently identified an additional member of the

complex bHLH protein network, Mist1 (Mak et al., 1996) The Mist1 gene exhibits a

complicated expression pattern, with transcripts being detected in several different

tissues including skeletal muscle derivatives (Lemercier et al., 1997) Specifically, Mist1

transcripts are expressed in skeletal muscle-forming regions during embryonic days E12.5and E16.5, but disappear rapidly as muscle cells differentiate This partially overlappingexpression pattern with the MyoD protein family prompted us to examine whether Mist1interacts with the muscle regulatory factors and whether it modulates MRF activity In thisreport, we show that Mist1 protein accumulates in undifferentiated myoblasts, but levelsrapidly decrease as cells begin to differentiate In addition, although Mist1 associates withMyoD, the heterodimer complex of Mist1–MyoD is unable to bind to muscle-specific E-boxelements Mist1 also inhibits MyoD from inducing muscle fiber formation in transfectedC3H10T1/2 fibroblasts and represses MyoD from activating gene transcription inmyogenic cells This repression mechanism is complex, however, involving DNA bindingcompetition as well as titration of E-proteins and the presence of a strong repressordomain in the N-terminus of the Mist1 protein Our results suggest that Mist1 inhibits theactivity of the muscle regulatory factors in skeletal muscle and also raise the possibilitythat Mist1 may serve as a regulator of other bHLH proteins in additional developmentalsystems

Mist1 protein is co-expressed with MyoD in proliferating myoblasts

We previously reported the cloning and characterization of a cDNA encoding a novel bHLHprotein, Mist1, which exhibits a unique expression pattern during embryonic development

and in adult tissues (Lemercier et al., 1997) Although Mist1 is capable of binding to an

E-box DNA target sequence as a homodimer, Mist1 protein complexes do not activate gene

transcription when tested in a variety of experimental systems (Lemercier et al., 1997) Since the Mist1 gene is expressed in skeletal muscle-forming regions during embryonic

development, we set out to investigate the possibility that Mist1 may regulate members

of the MyoD MRF family

To address this possibility, a derivative of the C2 myogenic cell line (C2C7) was tested forMist1 expression C2 cells express MyoD in undifferentiated myoblasts as well as in

differentiated myotubes (Vaidya et al., 1989) Examination of Mist1 protein accumulation

in these cells reveals that Mist1 is present at relatively high levels in undifferentiatedmyoblasts, but as differentiation proceeds over a 3 day period Mist1 protein accumulationbecomes reduced, with only very low levels, if any, detected in the myotube cultures(Figure 1) Conversely, as differentiation progresses, high levels of skeletal myosinaccumulate Thus, Mist1 is co-expressed with MyoD in myoblasts but not in differentiatedmyotubes, suggesting the possibility that Mist1 and MyoD may interact within a commonregulatory network in proliferating myogenic stem cell populations

Mist1 forms homodimers and heterodimers with MyoD and E-proteins

In order to examine potential interactions between MyoD and Mist1, electrophoretic

mobility shift assays were performed using in vitro translated proteins and a 32P-labeled box oligonucleotide As shown in Figure 2, Mist1 homodimers readily interact with the E-box sequence whereas MyoD homodimers do not bind to the DNA When MyoD and Mist1are co-incubated with the E-box probe, no new DNA-bound complexes are observed.Interestingly, addition of MyoD to these reactions produces a noticeable reduction inMist1 homodimer binding, suggesting that MyoD interacts with Mist1, but that the MyoD–Mist1 heterodimer complex does not bind to the target DNA Mist1 also forms aheterodimer with the E-protein E47, but in this instance the Mist1–E47 complex binds tothe E-box site (Figure 2)

E-The reduction in Mist1–Mist1 DNA complexes observed in the presence of MyoD prompted

us to investigate directly whether Mist1 interacts with this bHLH muscle regulatory factor

For these studies, in vitro protein–protein interaction assays were performed using a

histidine-tagged Mist1 protein (His-Mist1) and a variety of 35S-labeled in vitro translated

proteins Nickel beads, with or without His-Mist1, were incubated with the 35S-labeledproteins, and bound proteins were subsequently resolved on an SDS–PAGE gel When His-Mist1 is incubated with [35S]Mist1, a strong, specific 24 kDa band is detected byautoradiography (Figure 3), confirming that Mist1 homodimers readily form under theseconditions Similarly, Mist1 forms heterodimers with the [35S]MyoD and [35S]E47 proteins(as well as with [35S]MRF4, data not shown) However, no interactions are observed whenHis-Mist1 and [35S]c-Fos or [35S]MLP are tested These in vitro studies demonstrate that

Mist1 interacts with the bHLH proteins Mist1, MyoD, MRF4 and E47, and that formation ofhomodimers or heterodimers occurs independently of DNA target sites In addition, Mist1protein interactions are limited to bHLH factors since no interactions are observed with c-Fos (leucine zipper protein), MLP (LIM-domain protein) or with the MADS domain proteinMEF-2 (data not shown)

Mist1 functions as a potent inhibitor of MyoD activity

Given that the Mist1 and MyoD genes exhibit partial overlapping expression patterns (Lemercier et al., 1997) and the proteins directly interact with each other, we next

examined whether Mist1 influences MyoD activity when co-expressed in cells C3H10T1/2fibroblasts were transfected with MyoD, Mist1 or a combination of expression plasmids,and the number of MyoD-generated myocytes was measured As expected, MyoDefficiently induces the formation of muscle fibers that express skeletal muscle myosin(Figure 4A) However, when Mist1 is co-transfected with MyoD, fiber formation andmyosin expression are severely reduced (Figure 4A) Analysis of cell extracts preparedfrom these cultures confirmed that high levels of myosin protein accumulate in the MyoD-transfected cells but very little skeletal myosin is detected in cells co-transfected with theMyoD plus Mist1 cDNA constructs (Figure 4B) Importantly, the reduction in fiberformation and myosin accumulation is not due to a Mist1-dependent decrease in MyoDprotein levels Cells transfected with the MyoD expression plasmid, with or without theMist1 expression plasmid, exhibit similar levels of MyoD (Figure 4B) Thus, Mist1 inhibitsMyoD activity without altering the synthesis or stability of the MyoD protein

In an effort to identify the mechanism(s) by which Mist1 represses MyoD activity, we alsoexamined the effect Mist1 has on MyoD-dependent transcription For these studies, MyoDand/or Mist1 expression plasmids were co-transfected into C3H10T1/2 cells with troponin Iluciferase (TnI-Luc) or (E-box)4-CAT (containing four copies of an E-box site) reportergenes As expected, cells expressing MyoD generate high levels of TnI luciferase activity(Figure 5A) However, when MyoD and Mist1 expression plasmids are tested at a 1:1 ratio,TnI-Luc expression is repressed by 85% A similar Mist1-dependent repression (75%reduction) is obtained when the (E-box)4-CAT reporter gene is tested (Figure 5A) At higherMist1:MyoD ratios (5:1) (Figure 5B), TnI-Luc expression is completely absent, confirmingthat Mist1 functions as a potent inhibitor of MyoD activity Once again, this inhibition isspecific since Mist1 does not influence the transcriptional activity of non-bHLH factors nordoes Mist1 affect transcription from control SV40-Luc, CMV-LacZ and MEF2-CAT reportergenes (data not shown)

The repressor function of Mist1 is somewhat analogous to the repressor activityassociated with Id, a HLH protein that inhibits the activity of the MRFs by forming DNA-

binding inactive complexes of Id–MyoD or Id–E-proteins (Jen et al., 1992) In order to

compare the ability of Id and Mist1 to repress MyoD activity, we tested both proteins inMyoD transfection assays When equal concentrations of Id and MyoD expressionplasmids are tested in C3H10T1/2 cells, TnI-Luc activity is reduced 50%, and at a 5:1molar ratio, only 10% of MyoD-dependent activity remains (Figure 5B) In contrast, atthese concentrations, Mist1 represses TnI-Luc activity by 85 and 100%, respectively.Although it is difficult to compare directly the repressor activity associated with these twoproteins, it is clear that Mist1 inhibits MyoD activity at least as well as Id represses MyoDactivity Additionally, since Mist1 is a bHLH factor with DNA-binding activity, and Idrepresents a HLH factor lacking DNA-binding activity, the mechanism of Mist1-inducedinhibition may be substantially different from the mechanism by which Id represses theMRFs (see below)

Mist1 repression partially occurs through a DNA-binding mechanism

There are several mechanisms by which Mist1 may inhibit MyoD activity, including: (i)titration of E-proteins from MyoD, resulting in a decreased number of transcriptionallyactive MyoD–E-protein complexes, (ii) occupancy of essential E-box sites, therebypreventing MyoD–E-proteins from binding to DNA or (iii) dimerization with MyoD, forming

a DNA-binding inactive complex To understand further the contribution of proteindimerization versus DNA binding involved in Mist1 repression, C3H10T1/2 fibroblasts weretransfected with expression plasmids encoding MyoD, E47 or a tethered MyoD E47protein (Neuhold and Wold, 1993) in the presence or absence of Mist1 As shown

previously, Mist1 efficiently represses MyoD-dependent TnI-Luc expression, whereas the

related bHLH factor, Mash1 (Johnson et al., 1990), has little effect on MyoD activity

(Figure 6A) Co-transfection of MyoD and E47 into the C3H10T1/2 cells produces the samehigh levels of TnI-Luc activity as observed with MyoD alone Interestingly, inclusion ofMist1 in the MyoD + E47 group again generates an 85% reduction in reporter geneexpression, suggesting that the additional E47 protein is not sufficient to reverse theMist1-dependent inhibition of TnI-Luc gene expression Thus, it seems likely that Mist1repression is not dependent on a simple E-protein titration model As a confirmation ofthis hypothesis, we also co-transfected Mist1 with a tethered MyoD E47 construct(Neuhold and Wold, 1993) MyoD E47 possesses the same activating properties that areassociated with the individual MyoD and E47 proteins, but the tethered MyoD E47protein is highly resistant to Id repression since stable intramolecular MyoD–E47heterodimer complexes rapidly form (Neuhold and Wold, 1993) Surprisingly, when Mist1

is co-transfected with MyoD E47, a large decrease (90%) in reporter gene activity isobserved, which also parallels the decrease in MyoD E47-induced muscle fiber formation(Figure 6B) Conversely, overexpression of Id has little effect on MyoD E47-inducedmyogenesis (Figure 6B), reinforcing the idea that Mist1 repression does not involve astrict titration of available bHLH factors but primarily involves a DNA-binding mechanism

To test this latter hypothesis, electrophoretic mobility shift assays were performed using a

32P-labeled E-box probe and in vitro translated proteins Co-incubation of MyoD with E47

generates the predicted MyoD–E47 heterodimer complex that binds efficiently to the box site (Figure 7A) Incubation of Mist1 with E47 results in two distinct DNA-bindingcomplexes, a Mist1–Mist1 homodimer and a Mist1–E47 heterodimer When increasingamounts of Mist1 are added to the MyoD + E47 reactions, a progressive reduction in theformation of DNA-bound MyoD–E47 complexes occurs that coincides with an increase inthe formation of DNA-bound Mist1–E47 heterodimers and Mist1–Mist1 homodimers Theseresults suggest that Mist1 is able to disrupt a transcriptionally active complex, such asMyoD–E47, to form two additional DNA-binding complexes (Mist1–Mist1, Mist1–E47) that,none the less, remain transcriptionally inactive on muscle genes Interestingly, Mist1 has

E-no effect on the DNA-binding activity associated with the MyoD E47 tethered protein(Figure 7B), even though MyoD E47 transcriptional activity is severely repressed by Mist1(Figure 6A)

The experiments described above strongly support the idea that Mist1 represses skeletalmuscle differentiation primarily through occupancy of E-box sites and not throughtitration of specific bHLH factors in the cell In order to test this hypothesis, we generated

a mutant Mist1 cDNA (Figure 8A) that contains a substitution of the basic domain aminoacids RER with GGG (Mist1mut basic) (see Materials and methods for details) We reasonedthat an altered protein which retains dimerization but not DNA-binding activity would nolonger function as a repressor if Mist1 repression occurs primarily through DNAinteractions As predicted, both the wild-type Mist1 and the Mist1mut basic proteins efficiently

form homodimers when tested in an in vitro binding assay (Figure 8B) However, Mist1mut basic homodimers fail to bind to E-box sites whereas the wild-type Mist1 protein readilyforms a Mist1–Mist1–DNA complex (Figure 8C)

In order to test the new Mist1mut basic protein in vivo, we co-transfected Mist1 expression

plasmids along with a MyoD expression plasmid and the TnI-Luc reporter gene intoC3H10T1/2 fibroblasts As shown in Figure 8D, wild-type Mist1 inhibits MyoD-inducedactivation of the TnI-Luc reporter whereas Mist1mut basic has no effect, despite the fact thatthe Mist1mut basic protein stably accumulates in the cell and translocates to the nucleus

(data not shown) As a further test for in vivo activity, we examined the ability of Mist1 to

block the activity of the endogenous MyoD protein in the myogenic cell line C2C7 Aspredicted, C2C7 cells expressing the wild-type Mist1 protein fail to activate expression ofthe TnI-Luc reporter gene (Figure 8E) In contrast, the Mist1mut basic protein has no effect onC2C7 terminal differentiation, even though Mist1mut basic is capable of dimerizing with MyoDand E47 (unpublished data) Taken together, these data suggest that Mist1 primarilyinhibits MyoD in proliferating myoblasts through a DNA-binding mechanism and that thecontribution of dimerization with endogenous bHLH proteins is not a significant factor

given that the Mist1 protein, which retains dimerization function, no longerrepresses MyoD in these myogenic cells.

Mist1 possesses an N-terminus repressor domain

The ability of Mist1 to repress MyoD activity could involve a passive mechanism ofoccupancy of E-box sites or involve an active mechanism by which a repressor domaininhibits E-box-dependent transcription In order to distinguish between these possibilities,several Gal4 DNA-binding domain (Gal4 DB)–Mist1 fusion proteins were generated andtheir ability to repress expression of CAT reporter genes analyzed For these studies, we

took advantage of a CAT reporter system (Hollenberg et al., 1995) in which five

Gal4-binding sites are positioned downstream of eight LexA operators (L8G5-CAT) (Figure 9).This construct then was co-transfected into C3H10T1/2 cells with or without a LexA-VP16activator and various Gal4–Mist1 fusion constructs As predicted, when a plasmidcontaining only the Gal4 DB is tested, no CAT activity is detected, confirming that theGal4 DB has no transcriptional activity in this system (Figure 9) However, when the LexA-VP16 expression plasmid is co-transfected with Gal4 DB, high levels of CAT activity areobserved When Gal4 DB–Mist1(1–197) (full-length Mist1) is tested alone, again no CATactivity is detected, confirming our previous results that Mist1 lacks a functional

activation domain (Lemercier et al., 1997) Interestingly, when LexA-VP16 is included in

this group, the Gal4 DB–Mist1(1–197) represses LexA-VP16-induced CAT activity by 90%.Similar repression is also obtained when Gal4 DB–Mist1(1–126) and Gal4 DB–Mist1(1–71)are tested, whereas no repression is observed with Mist1(131–197) The repressionassociated with Gal4 DB–Mist1(1–197), –Mist1(1–126) and –Mist1(1–71) is dependent onDNA binding since no repression occurs with a reporter gene lacking Gal4-binding sites(L8-CAT) To examine if the repression domain of Mist1 can be overridden, we alsogenerated a Mist1 construct containing the VP16 transcriptional activation domain Asshown in Figure 9, the control Gal4 DB–VP16 efficiently activates expression of the L8G5-CAT reporter gene Interestingly, when the Gal4 DB–Mist1(1–197)–VP16 expressionconstruct is tested, high levels of CAT activity are also obtained Similarly, Gal4 DB–Mist1(1–197)–VP16 no longer represses the activity of LexA-VP16 (Figure 9) Takentogether, these results suggest that Mist1 inhibits gene transcription through a DNA-binding, active repression mechanism, with a potential repressor domain located at the N-terminus of the protein

Discussion

Skeletal myogenesis is controlled by a number of transcription factors, including the fourmembers of the MyoD family Forced expression of the myogenic bHLH proteins in non-muscle cells often overrides pre-existing developmental fates by instructing cells to enter

a myogenic lineage pathway The dominance of the MRFs underscores the need tomodulate their activity precisely in muscle-forming regions since overactivity may result

in inappropriate muscle formation (Weintraub et al., 1989; Miner et al., 1992) Similarly,

inhibition of MRF activity is critical to non-muscle tissues to ensure that the normaldevelopmental potential of these cells is not adversely affected In this study, we reportthat Mist1, a novel bHLH protein that is transiently expressed in embryonic skeletalmuscle-forming regions, interacts with MyoD and inhibits the transcriptional activityassociated with this MRF The mechanism by which Mist1 represses MyoD activity iscomplex, and involves a combination of occupancy of E-box sites as well as use of anactive repressor domain This combination defines a new repressor protein that probablyoperates to modulate activity of bHLH factors in myogenesis as well as bHLH factorsfound in a variety of other developmental systems

The need for negative regulators of MyoD activity is an obvious requirement when oneconsiders how myogenic stem cells remain as an undifferentiated cell population in tissueculture model systems, in cells found within the developing limb buds and in regeneratingsatellite cells In each case, proliferating myoblasts express high levels of MyoD and yetremain undifferentiated until appropriate environmental conditions are attained,conditions that are usually associated with the depletion of specific growth factors (e.g

fibroblast growth factors) from the local environment Indeed, the role of growth induced post-translational modifications associated with MyoD obtained from myoblastsversus differentiated myotubes has been examined extensively (reviewed in Ludolph andKonieczny, 1995) Many of these studies have failed to identify a specific modification(s)that is responsible for converting MyoD from an inactive to an active transcription factorcomplex A second avenue of investigation has been to identify factors that interactdirectly with MyoD to block its activity Candidate gene products are Id and Twist, both ofwhich prevent MyoD activity in a variety of experimental situations In this current study,

factor-we provide supporting data that Mist1 similarly may serve as a negative regulator ofMyoD The relevance of Mist1 acting as a MyoD negative regulator is even morecompelling given that the Mist1 protein accumulates to high levels in myoblasts, whereMyoD activity needs to be kept in check, and then Mist1 protein levels decrease indifferentiated myotubes, where MyoD activity is required to be maximal

Mist1 contains a central bHLH motif that is utilized for DNA binding to E-box regulatoryelements and for protein dimerization with other bHLH factors Unlike most bHLH factors,Mist1 lacks a transcription activation domain and instead exhibits an inhibitory potential.The mechanism of inhibition is complicated by the fact that Mist1 forms both homodimerand heterodimer bHLH complexes For instance, Mist1–MyoD heterodimers form, but sincethese complexes do not bind to DNA, a net decrease in the active MyoD protein poolensues In this instance, Mist1 resembles the HLH protein Id, which interacts with MyoD

and E47, forming inactive heterodimers that no longer bind to DNA (Benezra et al., 1990; Jen et al., 1992) However, unlike Id, Mist1 efficiently challenges a tethered MyoD E47

complex (Neuhold and Wold, 1993) when assayed in an E-box-dependent reporter genesystem or when examining MyoD E47-induced myofiber formation These results suggestthat Mist1 and Id utilize different molecular strategies for repressing MyoD activity This isespecially relevant for Mist1–E47 heterodimers, which continue to interact with E-boxsequences without leading to transcriptional activation Thus, although Mist1 formsinactive heterodimer complexes with other bHLH factors, we favor a model by whichMist1-directed repression occurs primarily through muscle-specific E-box occupancy byMist1–Mist1 homodimers (see Figure 10) In this scenario, Mist1 levels in proliferatingmyoblasts are sufficient to compete with MyoD for muscle-specific E-box sites When cells

commit to terminal differentiation, Mist1 transcription ceases, leading to functional MyoD

activity and activation of muscle-specific genes

Mist1 shares several similarities with the HES-1 proteins, the mammalian homologs for

Drosophila hairy and enhancer of split Like Mist1, HES-1 is able to repress MyoD-induced

fiber formation and inhibit expression of muscle-specific reporter genes, probably by

preventing MyoD from binding to E-box sequences (Sasai et al., 1992) In addition, HES-1

uses an N-box recognition motif (-CACNAG-) to trigger an active transcriptional repressionmechanism Although there are some similarities between Mist1 and HES-1, Mist1 cannot

be classified as a HES family member since it is unable to bind to N-box sequences(unpublished data) and Mist1 lacks the conserved C-terminal amino acid motif WRPW

which is present in hairy, enhancer of split and HES proteins and which is involved in protein dimerization and transcriptional repression (Dawson et al., 1995; Fisher et al., 1996) Other HLH proteins which inhibit muscle development are Id (Jen et al., 1992) and Twist (Rohwedel et al., 1995; Spicer et al., 1996; Hamamori et al., 1997) The mechanism

of myogenic inhibition associated with these factors often involves the disruption ortitration of molecules that participate in myogenesis, namely members of the MyoD andE-protein families However, these repressor proteins also interact with specific co-regulators of the MyoD family to inhibit muscle-specific transcription similarly Forinstance, Twist has been shown to form a heterodimer complex with E-proteins, which

then represses the ability of MEF2 transcription factors to promote myogenesis (Spicer et

al., 1996) Since MEF2 cooperates with MyoD to generate high levels of muscle

transcription, restricting MEF2 activity may ultimately modulate MyoD activity We havenot examined formally whether Mist1 blocks MyoD–MEF2 interactions However, given the

muscle and non-muscle expression patterns associated with the Mist1 and MEF2 genes, it

is tempting to speculate that Mist1 may alter MEF2 activity in other developmental

contexts by associating with additional bHLH proteins that may be part of differentregulatory pathways Further studies will be designed to examine these possibilities.

At this time, the mechanism by which Mist1 homodimers, when bound to specific E-boxsites, inhibit gene expression remains unknown Clearly, Mist1 utilizes an N-terminalrepressor domain that is sufficient to influence negatively transcription of heterologoussystems Thus, one possibility is that the Mist1 repression domain directly interacts with aspecific TATA box-binding protein (TBP)-associated factor (TAF) that is associated with TBP

to inhibit polymerase II transcription Similarly, Mist1 may function by recruitingdeacetylases to potential transcription complexes, thereby promoting core histonedeacetylation and transcriptional repression Analogous transcription mechanisms havebeen proposed for several well characterized transcription factor complexes, including N-

CoR, a known co-repressor for the thyroid hormone receptor (Hörlein et al., 1995; Zamir

et al., 1996; Heinzel et al., 1997) and mSin3, a co-repressor for the Mad–Max transcription

complex (Ayer et al., 1995; Alland et al., 1997) Additional function/structure studies will

be required to characterize the essential properties associated with the Mist1 repressordomain Regardless of the precise mechanism, however, the specificity of the system

relies on two variables; the embryonic expression pattern of the Mist1 gene and specific

E-box-dependent target genes

Materials and methods

DNA constructions

All of the cDNAs tested in mammalian cells were subcloned into the pcDNA3 expressionvector (Invitrogen) MyoD, E47 and MyoD E47 cDNA inserts were isolated from pECE-MyoD, pECE-E47 and pECE-MyoD E47, respectively (Neuhold and Wold, 1993) by

digestion with HindIII and EcoRI and subcloned into pcDNA3 at the HindIII and EcoRI sites The Mash1 cDNA was excised from pBS-Mash1 (Johnson et al., 1990) using a KpnI digest and ligated into the KpnI site in pcDNA3 The mouse Id1 cDNA was digested from pE:Id (Benezra et al., 1990) with SmaI and the isolated insert ligated into the EcoRV site in

pcDNA3 In order to produce Gal4 DB fusion proteins, the Gal4 DB region, plus part of the

polylinker from the Gal4 (1–147) plasmid (Sadowski et al., 1988), was digested with

HindIII and XbaI and ligated into pcDNA3 Point mutations were introduced into

pcDNA3-Mist1 using the 'Quick Change Site Directed Mutagenesis' kit from Stratagene CloningSystems The Mist1 basic mutant (Mist1mut basic) contains alterations in amino acids 80–82from RER to GGG Mutations were confirmed by direct DNA sequencing

All Mist1 deletions were generated by PCR as described previously (Lemercier et al.,

1997) using pcDNA3-Mist1 (HA epitope-tagged at the N-terminus) as a template and the

following 5' primers containing an EcoRI site and 3' primers containing a XbaI site: Mist1(1–197), 5' primer 5'-GGAATTCATGTATCCTTATGACGT-3' (EcoRI site underlined), and 3' primer 5'-CGTCTAGAGCTCCCCTCTCTGAAG-3' (XbaI site underlined); Mist1(1–126), 5'

primer as above, and 3' primer 5'-CGTCTAGAGGCTGTCAGCGACTTG-3'; Mist1(1–71), 5'primer as above, 3' and primer 5'-CGTCTAGAACGCTGTTCTCCCTG-3'; Mist1(131–197), 5'primer 5'-GGAATTCATGTCCAGCAGCCGCCT-3', and 3' primer 5'-

CGTCTAGATTACCAGTCTGGG GCT-3' All PCR products were digested with EcoRI and XbaI and then ligated in-frame into the pcDNA3-Gal4 DB plasmid at the EcoRI and XbaI sites.

pcDNA3-Gal4-VP16 was obtained by isolating the Gal4–VP16 insert from a Gal4–VP16

plasmid (Sadowski et al., 1988) using HindIII and XbaI and subsequently subcloning the

insert into the pcDNA3 vector Gal4 DB–Mist1–VP16 was generated by ligation of Gal4 DB–

Mist1 (1–197) into pcDNA3-VP16 (Lemercier et al., 1997) using the available HindIII and

XbaI sites Bacterial His-tagged Mist1 protein was obtained by digesting the Mist1 cDNA

from pcDNA3-Mist1 with KpnI and XhoI and ligating the 700 bp fragment in-frame into pQE31 (Qiagen, Inc.) at the available KpnI and SalI sites.

Cell culture and DNA transfections

The C2C7 mouse myoblast line was derived from the original C2C12 myogenic cell line

(Blau et al., 1983) and was obtained from Anne Fernandez (Montpellier, France) C2C7

myoblasts were maintained in Dulbecco's modified Eagle's medium (DMEM):Ham F12nutrient mix (1:1) supplemented with 10% fetal bovine serum plus penicillin (100 U/ml)and streptomycin (100 g/ml) Terminal differentiation into myotubes was induced byreplacing the culture medium with high glucose DMEM containing 2% horse serum for 1–3days

For cell transfections, C3H10T1/2 fibroblasts were plated at a density of 1 105 cells/35

mm well in basal Eagle's medium (Gibco) containing 10% fetal bovine serum pluspenicillin (100 U/ml) and streptomycin (100 g/ml) C2C7 myoblasts were similarly plated

in DMEM:Ham F12 nutrient mix growth medium as described above Luciferase or CATassays involved transfecting cells with 0.2 g of each test plasmid (unless otherwisestated) along with 1 g of a troponin I luciferase or 2 g of (E-box)4-CAT reporter plasmid

as described previously (Johnson et al., 1996) When pcDNA3-MyoD was present in the

DNA precipitate, cells were induced to differentiate in low glucose DMEM containing 2%horse serum After 48–72 h, luciferase activities were measured using a luciferase kit

(Luciferase Assay System, Promega) as described previously (Lemercier et al., 1997).

Luciferase activities were normalized to the protein content of each sample For LexA-CATassays, 2 105 C3H10T1/2 fibroblasts/60 mm dish were transfected with 2 g of each Gal4

DB fusion protein plasmid, 2 g of L8G5-CAT reporter plasmid (eight LexA operators withfive Gal4-binding sites, upstream of the CAT gene) or 2 g of the control reporter L8-CAT(eight LexA operators upstream of the CAT gene), in the presence or absence of 0.4 g of

LexA-VP16 activator plasmid (Hollenberg et al., 1995) After 72 h, cell extracts were prepared (Naidu et al., 1995) and protein content determined The amount of cell extract

used for each CAT assay was normalized to the protein level of each sample

Immunohistochemistry

C3H10T1/2 cells (5 105 cells/100 mm dish) were transfected as described above with 10

g of the pcDNA3-Mist1 expression vector with or without 10 g of pcDNA3-MyoD DNAconcentrations were normalized using the empty pcDNA3 vector as a control After 72 h

in differentiation medium, cells were rinsed three times in phosphate-buffered saline(PBS) and then fixed in a 20:2:1 solution of 70% ethanol/formalin/acetic acid for 1 min at

4°C as described by Kong et al (1995) Following a 30 min permeabilization/blocking step

(10 mM Tris pH 8.0, 150 mM NaCl, 0.1% NP-40, 5% horse serum), the plates wereincubated with an MF-20 hybridoma supernatant (specific for striated muscle myosinheavy chains) for 1 h under gentle shaking and then rinsed three times in PBS Theprimary complexes were detected using a biotinylated anti-mouse antibody and ahorseradish peroxidase (HRP)–strepavidin conjugate (Vector Laboratories) Specificimmunocomplexes were visualized by DAB staining and examined under an Olympusinverted phase contrast microscope

Western blots

C3H10T1/2 fibroblasts were transfected as described above and then cell extracts wereprepared by lysing the cells in 25 l of 4 SDS–PAGE loading dye (250 mM Tris pH 6.8, 8%SDS, 40% glycerol, 4% -mercaptoethanol) Extracts were sonicated for 20 s and kept at-80°C until analyzed further Extracts were prepared similarly from myoblast and myotubecultures of C2C7 cells SDS–PAGE and Western blots were performed according tostandard protocols Blots were saturated for 1 h with TBS (10 mM Tris pH 8.0, 150 mMNaCl) containing 10% non-fat dry milk (w/v) prior to incubation with the primary antibody.For myosin detection, the blots were incubated in the MF-20 supernatant diluted 1:20 inTBS–10% non-fat dry milk for 1 h and then rinsed for 20 min in TBS–0.1% Tween-20 (TBS-T) The immunocomplexes were detected using an anti-mouse IgG–HRP conjugate (SantaCruz, Inc.) and a chemiluminescence detection kit (Amersham) For MyoD and Mist1detection, the primary antibodies were as follows: rabbit polyclonal anti-MyoD C20 (SantaCruz, Inc.), mouse monoclonal anti-HA epitope (Boehringer Mannheim) and rabbit

polyclonal anti-Mist1 serum raised against a GST–Mist1 fusion protein (Lemercier et al.,

1997) The immunocomplexes were detected as above using an anti-mouse or anti-rabbitIgG–HRP conjugate and chemiluminescence

Electrophoretic mobility shift assays

The DNA probes used in the gel mobility shift assay experiments were oligonucleotidesderived from the troponin I IRE E-box consisting of the wild-type sequence 5'-

GATCCGTCTGAGGAGACAGCTGCAGCTCC-3' (E-box site underlined) In vitro translated proteins were prepared as previously described (Lin et al., 1991; Lemercier et al., 1997)

using the pcDNA3 plasmids Typical binding reactions contained 5 l of 5 binding buffer[25 mM HEPES, pH 7.9, 50 mM KCl, 0.5 mM EDTA, 5 mM MgCl2, 5 mM dithiothreitol (DTT)

and 10% glycerol], 0.5 g of poly(dI–dC), 3 l of in vitro translated protein, 300 fmol of the

labeled DNA probe (typically 10 000 c.p.m.) in a total volume of 25 l The entirebinding reaction mixture was allowed to assemble at room temperature for 30 min Forthe Mist1 titration experiments, increasing amounts of Mist1 ranging from 1 to 4 involume were added to the binding reactions during the incubation Following incubation,all reactions were analyzed by loading the entire mixture onto a 5% non-denaturing geland electrophoresis at 100 V for 3 h at room temperature After electrophoresis, the gelwas dried at 80°C for 1 h and the bound complexes visualized by autoradiography

In vitro protein-binding assays

Approximately 1 g of bacterial produced His-tagged Mist1 fusion protein was conjugated

to nickel (Ni2+) agarose beads and mixed with 5 l of [35S]Mist1, -Mist1mut basic, -MyoD, -E47,

-MRF4, -c-Fos (Rauscher et al., 1988) or -MLP (Arber et al., 1994; Kong et al., 1997) and

allowed to incubate in 250 l of binding buffer (10 mM Tris–HCl pH 7.5, 200 mM NaCl,0.25% NP-40, 20 mM imidazole) at 4°C for 2 h with gentle rocking Following thisincubation, the complexes were subjected to three 1 ml washes with binding bufferfollowed by two additional washes with binding buffer minus NP-40 As a negative control,the 35S-labeled proteins were incubated with nickel beads alone (no His-tagged Mist1) inthe binding reaction The isolated pellets were heated to 95°C for 5 min in SDS samplebuffer and subjected to 12% SDS–PAGE gel electrophoresis Protein interactions werevisualized by autoradiography

Acknowledgements

We acknowledge the important technical contributions of Lea Longcor and DebbyBartolucci and are indebted to S.Hollenberg, D.Anderson and B.Wold for generouslysupplying some of the plasmids used in this study In addition, we thank Anne Fernandez(Montpellier, France) for providing the C2C7 cell line and Joel Gaffe, Sally Johnson andmembers of the Konieczny laboratory for critical comments on this work This work was

supported by grants to S.F.K from the National Institutes of Health and the AmericanHeart Association C.L was supported by an American Heart Association (Indiana Affiliate)Postdoctoral Fellowship and R.A.C by a Sloan Foundation Minority Graduate StudentFellowship.

Benezra R, Davis RL, Lockshon D, Turner DL and Weintraub H (1990) The protein Id: a negative regulator of helix–loop–helix DNA-binding proteins Cell, 61, 49–59

Blackwell TK and Weintraub H (1990) Differences and similarities in DNA-binding

preferences of MyoD and E2A protein complexes revealed by binding site selection Science, 250, 1104–1110

Blau HM, Chiu C-P and Webster C (1983) Cytoplasmic activation of human nuclear genes

in stable heterokaryons Cell, 32, 1171–1180

Buckingham ME (1994) Muscle: the regulation of myogenesis Curr Opin Genet Dev, 4, 745–751

Dawson SR, Turner DL, Weintraub H and Parkhurst SM (1995) Specificity of the

Hairy/Enhancer of split basic helix–loop–helix (bHLH) proteins maps outside of the bHLH domain and suggests two separable modes of transcriptional repression Mol Cell Biol, 15,6923–6931

Fisher AL, Ohsako S and Caudy M (1996) The WRPW motif of the hairy-related basic helix–loop–helix repressor proteins acts as a 4-amino-acid transcription repression and protein–protein interaction domain Mol Cell Biol, 16, 2670–2677

Hamamori Y, Wu H-Y, Sartorelli V and Kedes L (1997) The basic domain of myogenic basic helix–loop–helix (bHLH) proteins is the novel target for direct inhibition by another bHLH protein, Twist Mol Cell Biol, 17, 6563–6573

Heinzel T et al (1997) A complex containing N-CoR, mSin3 and histone deacetylase

mediates transcriptional repression Nature, 387, 43–48

Hollenberg SM, Sternglanz R, Cheng PF and Weintraub H (1995) Identification of a new family of tissue-specific basic helix–loop–helix proteins with a two-hybrid system Mol Cell Biol, 15, 3813–3822

Hörlein AJ et al (1995) Ligand-independent repression by the thyroid hormone receptor

mediated by a nuclear co-repressor Nature, 377, 397–404

Hu JS, Olson EN and Kingston RE (1992) HEB, a helix–loop–helix protein related to E2A and ITF2 that can modulate the DNA-binding ability of myogenic regulatory factors Mol Cell Biol, 12, 1031–1042